IDENTIFICATION OF DEMAND THROUGH STATISTICAL DISTRIBUTION MODELING FOR IMPROVED DEMAND FORECASTING.

IDENTIFICATION OF DEMAND THROUGH

STATISTICAL DISTRIBUTION MODELING FOR IMPROVED

DEMAND FORECASTING

ANGGOTA KELOMPOK • DEDI ISKANDAR Z. (115 100 307 113

012)

• ENDRA CAHYONO (115 100 307 113 011)

• FATWATUL AMALIA (115 100 307 113 009)

• LEEANANDA GALANG (115 100 307 113 013)

• MARINAYATI S ( 115 100 307 113 010 )

• Demand forecasting is an important aspect of business operation.

• It is applicable to many different functional areas such as sales, marketing and inventory management.

• Proper demand forecasting also allows for more efficient and responsive business planning.

• Tourism and manufacturing are the two major industries who adopt a wide range of demand forecasting and variability management solutions.

• Demand functions for goods are generally cyclical in nature with characteristics such as trend or stochasticity.

• Most existing demand forecasting techniques in literature are designed to manage and forecast this type of demand functions.

• In this paper, will be explained three types of demand functions and their mathematical representations.

• Demand data will simulate using the mathematical representations and model the simulated data to identify the statistical distributions.

• As such, would have established the relationship between demand type and statistical distribution of demand data.

Types of Demand Functions• TYPE 1

The first type of demand function is the generic cyclical model with trend. This type of demand function can be generalized into the following form as Equation (1).

Let,

- Yi = demand of product at time i

- Ti = upward or downward trend component of demand at time i

- Cij = cyclical component of type j at time i, where j = 1 to J

• TYPE 2

The second type of demand function is commonly known as stochastic demand. this type of demand function can be generalized into the following mathematical form as Equation (2).

dimana, Yi = demand level at time i

• TYPE 3

The last type of demand function is the lumpy demand function. This type of demand resembles stochastic processes but has its own unique characteristics. Three main characteristics were summarized from the literature.

• Variable: Fluctuations are present and related to some common factors (Wemmerlöv, 1986; Ho, 1995; Syntetos and Boylan, 2005).

• Sporadic: Demand can be non-existent for many periods in history (Ward, 1978; Williams, 1982; Fildes and Beard, 1992; Vereecke and Verstraeten, 1994; Syntetos and Boylan, 2005)

• Nervous: Each successive observations is different which implies low cross time correlation (Wemmerlöv and Whybark,1984; Ho, 1995; Bartezzaghi and Verganti, 1995).

• A lumpy demand distribution is defined as a demand which is extremely irregular with high level of volatility coupled with extensive periods of zero demand (Gutierrez, 2004).

• Y = F(I) at I = i

• Let,

- Yi = the level of a normal lumpy demand for time i

- Zi = the level of a modified lumpy demand for time i

- F(i) = the distribution which created the lumpy demand at time i

- F’(i) = distribution F(i) shifted by a constant value A

- A = fixed base level demand > 0

• Adding a fixed base level demand A to Yi to get Zi,

Z = Yi + A at I = i

Thus, Z = F´(I) at I = i

Application of Methodology on Real Case

• In this paper, given an example of application of the methodology in a real case

• We apply our methodology on a real case with demand data from a retailer specializing in luxury watches. The retailer is currently facing problems with forecasting the demand for luxury watches in several countries.

From Table 2, we can observe that Lumpy demand is best predicted by Holts Winter Additive model as opposed to Stepwise Auto-Regressive model with the lowest average MSE.

Conclusion• There are three characteristic types

of demand functions and using simulated data to establish the relationship between demand function and statistical distributions.

• In a real case, we can choose the best demand forecasting method what have a little error score.

THANK YOU……….